Friday, August 03, 2012

Memristors and Claim Construction

It is no surprise to regular readers of this blog that I oppose the hype surrounding HP's memristor. A recent online article in Wired covers some of my concerns (link).

However, even for those who support the memristor hype I would recommend caution when using the term "memristor" in the claims of patent applications. The original definition of the memristor from Chua required a non-linear relationship between magnetic flux linkage and electric charge. HP's definition from their 2008 article in Nature suggested that thin films of metal oxides may be considered memristors if there was a linear relationship between the drift of oxygen vacancies in the thin film and an applied current (or voltage). The problem is that this linearity is not true for most resistance switching materials. If an inventor or corporation uses the term "memristor" to limit their claims they face two potential problems if they try to enforce their patent.

1) The claims may be held invalid under 35 USC 112 for lack of enablement if the materials taught in the specification do not support linearity between oxygen vacancy (or ion) drift and the applied current (or voltage).

2) During a Markman hearing the claims may be limited to only a specific type of material (such as TiO2 as described by HP) even though many other resistance switching materials may be used in the invention.

So for those who choose to use the "memristor" term in their patent claims it is a much better strategy to reserve the term for a dependent claim while using more generic terminology for independent claims (e.g. memory resistor, programmable resistor, etc.) Also, it would be advisable to include alternatives in the specification based on the known types of materials exhibiting memory resistance effects (chalcogenides, perovskites, etc.) rather than limiting to the specific memristor term.